Explaining the OECD Wage Slowdown:

Recession or Labor Decline?

1

Bruce Western

Kieran Healy

Princeton University

March, 1998

1

Department of So ciology, 2-N-2 Green Hall, Princeton University, Princeton

NJ 08544-1010; [email protected]. Research for this pap er was supp orted

by grant SBR95-11473 from the National Science Foundation. Thanks to Paul

DiMaggio, Alex Hicks, Mike Hout, Paul Pierson, and Michael Wallerstein who

provided useful comments on a earlier version of this pap er.

Abstract

Wage growth slowed signi cantly in OECD countries in the 1980s and 1990s.

Market explanations trace the wage slowdown to a recession characterized

by in ationary sho cks, high unemployment, and slow pro ductivity growth.

Institutional accounts fo cus on the e ects of union density, collective bar-

gaining centralization, and lab or government. Analysis of time series from

18 countries for 1966 to 1992 yields some evidence for b oth theories b etween

1966 and 1974. Bayesian metho ds indicate a structural break in the wage

growth pro cess, linking the wage slowdown of the 1980s to the declining

power of lab or movements.

The pace of wage growth varies greatly across the advanced capitalist

countries. In the late 1960s and early 1970s, Europ ean wages rose by ab out

4or5eachyear. U.S. wages grew ab out a third as fast over the same p e-

rio d. In the 1980s and 1990s, real wage growth slowed dramatically through-

out Europ e and North America. Europ ean wage stagnation was asso ciated

with high unemployment and rising pressure on the welfare state OECD

1994; McFate 1995. In the United States, falling wages accompanied rising

poverty and inequalityFreeman 1995; Harrison and Bluestone 1988. The

recentwage slowdown thus forms part of a protracted decline in lab or market

p erformance that marks the end of a golden age of rising living standards

and rapid economic growth in OECD countries Glyn 1995; Harrison and

Bluestone 1988.

The OECD wage slowdown challenges institutional theories of economic

pro cesses. If institutions explain cross-national di erence, whywas wage

growth halted at similar times across so many institutional contexts? Market

explanations that minimize institutional in uences may b e more promising.

Such explanations emphasize the e ects of pro ductivity growth, unemploy-

ment, and in ation Chan-Lee, Co e, and Prywes 1987; OECD 1997. For

market explanations, the rst oil crisis signalled the b eginning of a sustained

p erio d of slow economic growth which arrested the rise in wages.

We prop ose an alternative account that fo cuses on the institutionalized

power resources of organized lab or movements. The key idea is that wages de-

p end on industrial relations and p olitical institutions that shap e and advance

workers' interests e.g., Korpi 1983; Goldthorp e 1984; Scharpf 1991. Union-

ization, collective bargaining centralization, and lab or governments each o er

organized lab or movements collective control over wage growth. To explain 1

the wage slowdown, we draw on research that do cuments the fall in union

density, bargaining decentralization, and a rightward shift in the p olicies

of pro-lab or parties Western 1997; Katz 1993; Baglioni and Crouch 1990;

Piven 1991; Hub er and Stephens forthcoming. These developments have

weakened union in uence on lab or market outcomes, causing a general slow-

down in wage growth.

Toinvestigate these ideas, we analyze newly available data and o er a

novel approach to studying institutional change. To analyze wage movements

we construct a complete series of real wage trends, combining OECD and

national sources from 18 countries b etween 1966 and 1992. The impact of

unions is measured using new time-series data from Visser 1996 and Golden,

Lange, and Wallerstein 1997. Finally,we presenta Bayesian metho d for

estimating the e ects of institutional change on wages. This metho d yields

clear evidence for a structural break in the pro cess of wage determination.

Wage Growth in 18 OECD Countries, 1966{1992

Table 1 summarizes trends in manufacturing sector wages that form the de-

p endentvariable for this pap er. These data show the p ercentage growth in

real hourly wage rates in manufacturing industries. Despite our fo cus on

manufacturing wages, these data provide a go o d indication of wage move-

ments in a country as a whole. For instance|in the United Sates, where

detailed industry-level data are available|annual growth in hourly manu-

facturing wages correlates at .96 with wage growth in private sector services,

at .95 with wage growth in transp ortation, and at .91 with wage growth in

all service industries. In general, wage levels vary greatly across industries,

countries, and demographic groups, but wage trends are signi cantly more 2

homogeneous OECD 1997, 6{11. Because wages are the largest comp onent

of household income, wage growth provides an imp ortant indicator of trends

in living standards.

From the mid-1960s to the 1970s, pay rises accelerated in Europ e. Spurred

by rapid in ation and the strikewaves of 1968{1969 , Europ ean wages climbed

by ab out 5 annually. An unprecedented economic b o om in Japan help ed

drive the fastest rate of wage growth in the OECD|an extraordinary annual

rate of around 8. Wages rose more slowly in the English-sp eaking countries.

Real earnings grew by ab out 3 a year in Canada and Britain. U.S. wage

growth was sluggish, averaging just under 1.5 a year.

The OPEC oil sho ck of 1973{1974 op ened a new p erio d of p o or lab or

market p erformance. Where wage growth was slow, real wage declines re-

placed the small pay rises of the previous decade. In the United States,

estimates show that real wages for workers from nearly all education and ex-

p erience groups fell b etween 1979 and 1991 Katz, Loveman and Blanch ower

1995, 33. In Europ e, the rapid pace of wage growth slowed signi cantly.In

France, Germany, and Italy,wage growth was halved in the 1980s and 1990s

compared to the p erio d b efore the oil crisis. How can we explain the general

stagnation of the rise in living standards in the advanced capitalist countries?

Market Forces and Wage Growth

Market explanations app ear to provide a convincing and general accountof

wage trends in OECD countries. These explanations fo cus on the forces of

supply and demand. Lab or demand is re ected in the unemployment rate.

Lab or supply is measured by pro ductivity growth and in ation.

When lab or demand is weak and unemployment is high, wage growth 3

Table 1. Summary of annual p ercentage growth in real hourly manufacturing wage

rates, 18 OECD countries.

1966{1973 1974{1982 1983-1992

Australia 2.14 1.30 -1.53

Austria 5.31 2.61 1.98

Belgium 5.91 2.90 .29

Canada 3.34 1.52 -.11

Denmark 5.75 1.70 1.15

Finland 5.55 1.17 2.19

France 5.06 3.11 .68

Germany 4.56 1.56 2.08

Ireland 6.32 3.77 .83

Italy 5.70 2.41 1.83

Japan 9.05 1.38 1.50

Netherlands 3.70 .94 .70

New Zealand 2.86 -.41 -2.05

Norway 3.79 2.00 1.73

Sweden 4.19 .22 .83

Switzerland 1.74 .67 .87

United Kingdom 3.15 1.01 2.76

United States 1.35 -.49 -.74

Average 4.38 1.45 .85

Source: See App endix 1. 4

slows. In comp etitive lab or markets, unemployment restricts wage increases

by raising comp etition among workers for scarce jobs. In unionized lab or

markets, workers are prevented from under-bidding union wages. In this

situation, unemployment increases the threat of lay-o s or business failures

and unions bargain less aggressively as a result. The negative relationship

between real wage growth and unemployment has thus b een observed in b oth

unorganized and highly unionized lab or markets OECD 1997, ch1;Volgy,

Schwarz, and Inwalle 1996; OECD 1994, 3{4; Layard, Nickell, and Jackman

1991, ch. 9.

The quality of the lab or supply also in uences earnings. For neo clas-

sical theory, market comp etition ensures workers are paid their marginal

pro duct; pro ductivity growth thus drives wage growth Hicks 1963, 8. The

assumption of comp etitive markets is not vital however. In unionized lab or

markets, employers nance wage rises out of pro ductivity gains, and pro-

ductivity increases are often written into union contracts Flanagan et al.

1983. Comparative studies of p ostwar time series of OECD countries thus

nd that sustained wage growth has dep ended on continuous improvements

in pro ductivity OECD 1997, 22; cf. Volgy et al. 1996.

The lab or supply is also in uenced by in ation. In neo classical theory,

workers supply a quantity of lab or in return for a certain real wage. Unex-

p ected in ation causes workers to overestimate the value of their wages and

oversupply their lab or as a result Friedman 1968, 7{11. Excess lab or supply

then drives down wages. In an alternativeinterpretation, in ation reduces

real wages b ecause employment contracts sp ecify nominal rather than real

quantities Keynes [1935] 1964, ch. 2. Often in ationary exp ectations are

built into union contracts in the form of cost of living adjustments. Even 5

in these cases, unexp ected in ation can reduce wages. Simple mo dels of

in ationary exp ectations involving rst di erences consistently show the de-

p endence of wages on price movements Layard et al. 1991, ch 9; OECD

1997; Chan-Lee et al. 1987; Volgy et al. 1996.

Supply and demand trends t the main pattern of variation in wage

growth. Following the rst oil sho ck, economic growth slowed throughout

all OECD countries. The purp orted causes of slow growth include low rates

of investment, de ationary p olicy, the globalization of nancial markets, and

the growth of service sector employment e.g., Glyn 1995; Epstein and Schor

1992; Gershuny 1983. Whatever the precise causes, the e ects app ear clear:

a general downturn in pro ductivity growth; a secular rise in unemployment,

particularly severe in Europ e; and a p erio d relatively high in ation in the

1980s. The fall in pro ductivity growth constricted pay rises. Rising unem-

ployment shifted the balance of market p ower from wage-earners to employ-

ers. Finally, the purchasing p ower of wages was dissolved by price sho cks and

p ersistent in ation. The generality of economic explanations is suggested by

econometric research which nds that p olitical factors add little explanatory

power and slowwage growth in the 1980s dep ends mostly on in ation and un-

employment trends Chan-Lee et al. 1987; OECD 1997, ch. 1; cf. O'Connell

1994.

Institutions and Wage Growth

Market forces only partly explain wage trends, b ecause earnings are also

in uenced by the institutionalized p ower resources of lab or movements. Two

theories describ e the impact of organized lab or on wages. The rst claims

that unions use their organizational p ower to b o ost wages through collective 6

bargaining. In the second, centralized unions and lab or governments restrain

wage growth in return for low unemployment.

Industrial Relations and Labor Government

The p ositive e ect of unions on wages is basic to p ower resource theories of

organized lab or O'Connell 1994; Cohn 1993; Rubin 1986. In this approach

workers and employers have con icting interests in wage growth. Workers

want rapid wage growth to expand their incomes; employers want slowwage

growth to control pro duction costs. Armed with the threat of strikes, unions

use their organizational strength to push for higher wages. The p ositive e ect

of unions on wages thus dep ends on the level of union organization, or union

density|the p ercentage of unionized workers in the work force. Positive

unionization e ects have b een found in a wide variety of contexts, in studies

of micro data on the union wage premium and in comparative researchon

aggregate wage e ects Blanch ower and Freeman 1992; O'Connell 1994.

With centralized collective bargaining, strong unions may restrain rather

than fuel wage growth Olson 1982; Crouch 1985; Calmfors and Drill 1988.

Centralized bargaining transforms con ict b etween unions and employers into

co op eration. In this theory, the costs of wage growth in terms of unemploy-

ment are distributed throughout the economy. Under centralized bargaining,

union leaders represent the entire lab or force in national wage talks. Con-

sequently, the unemployment e ects of wage claims are directly exp erienced

by the union's constituency. Centralized unions thus have an incentive to re-

strain wage claims. With decentralized bargaining, union pay rises narrowly

b ene t workers in a given plant or rm, but the costs are distributed over

the whole lab or market. The resulting free-rider problem provides individual 7

unions with little incentive to restrain militancy. Empirical studies thus nd

evidence of wage restraint in countries with national or sectoral bargaining

Layard et al. 1991, ch. 9.

Political parties may also contribute to wage restraint Crouch 1985, 109;

Headey 1970. Lab or governments assist wage restraintby o ering tax and

so cial welfare guarantees to unions. These p olicies maintain incomes while

unions restrain their bargaining p ower. Conservativegovernments have less

co op erative relations with unions, obstructing a co ordinated approachto

economic p olicy. Conservatives may also b e less inclined to re ationary mea-

sures like so cial sp ending. Previous researchthus asso ciates the combined

impact of so cial demo cratic parties and centralized unions with low unem-

ployment and strong economic growth Scharpf 1991; Alvarez, Garrett, and

Lange 1991; Hicks 1994.

The current fo cus on institutional features of lab or movements may ne-

glect lab or militancy as a p otential source of wage growth. Although other

studies examined the e ects of strikes, they fo cused on the exp eriences of in-

dividual countries Rubin 1986; Hibbs 1987; Cohn 1993. The utility of strike

e ects for comparative analysis seems doubtful b ecause real wage growth has

b een slowest in the Anglo countries where strike activity has b een greatest.

High wage growth countries like Austria, Germany, and Japan, havevery low

strike rates Korpi and Shalev 1979. Comparative analysis of strike e ects

is also hamp ered by missing data for some countries.

Labor Decline

Recent research claims that these indicators of lab or institutionalization|

union density, bargaining centralization, and lab or government|play a dif- 8

ferent role in contemp orary lab or markets compared to the 1960s and early

1970s e.g., Streeck 1993; Hub er and Stephens forthcoming; Visser 1992a.

The changing impact of institutions is linked to three main developments:

1 the declining bargaining p ower of unions, 2 the growth of lo cal wage

bargaining, and 3 the rightward shift of left-wing parties.

Unions lost p ower in the 1980s and 1990s due to organizational decline

and growing p olitical and economic adversity. Union density fell in nearly

all OECD countries in the 1980s Western 1997. As a result, the p ower

of the strikewas severely weakened, and lab or militancy fell sharply Shalev

1992. In some countries|notably United States and the United Kingdom|

density decline was linked to anti-union campaigns by conservativegovern-

ments Western 1997. These campaigns also defused aggressivewage bar-

gaining byweakening legal protections for strikes Weiler 1990; Marsh 1992.

Economic conditions also diminished bargaining p ower. Persistent high un-

employment in Europ e constrained union wage claims. The growth of for-

eign trade placed unionized lab or in the North in comp etition with low-wage

exp orters of the South Wo o d 1994. In addition to organizational losses,

unions thus lost bargaining p ower through p olitical attacks on industrial

action, recession, and new comp etitive pressures from abroad. These devel-

opments havetwo empirical implications. First, the e ect of union densityon

wage growth may b e unchanged, but wages fall b ecause union density falls.

Second, declining bargaining p ower will likely change the e ect of union den-

sityonwage growth. Due to lost bargaining p ower, even unions that maintain

organization may b e unable to sustain wage growth in the 1980s and 1990s.

In short, the p ositive e ect of unions on wages is likely to decline in the last

two decades. 9

While the pursuit of their sectional interest in higher wages was weakened,

unions also met new obstacles to their general interest in wage restraint. Due

to the growth of rm-level bargaining in OECD countries during the 1980s,

central wage agreements decreasingly in uenced aggregate wage trends. The

extent of bargaining decentralization is disputed by comparative researchers.

Lange, Wallerstein and Golden 1995; Golden, Lange, and Wallerstein 1993

nd signi cant continuity in collective bargaining through the 1970s and

1980s in a sample of 16 OECD countries. Still, other research nds that

lo cal bargaining has ourished alongside centralized wage talks Katz 1993;

Baglioni and Crouch 1990. In Norway, for example, national bargaining

was uninterrupted during the 1980s, but lo cal wage drift consumed 80 of

all wage increases by the mid-1980s Mo ene and Wallerstein 1993. Reg-

ular industry-level wage rounds were also conducted in Germany through

the 1980s, but rm-level wage bargaining in works councils also expanded

signi cantly during this p erio d Thelen 1993. In short, the in uence of

centralized bargaining mayhave ero ded through the 1980s and 1990s, even

where formal measures of bargaining centralization show little change. As a

consequence, the negative e ects of centralized bargaining on wage growth

has likely declined in the 1980s and 1990s.

In addition to organized lab or's ero ded industrial p osition, comparative

researchers also p oint to a \decline of so cial demo cracy." While many leftist

parties were in opp osition through the 1980s Pontusson 1995; Piven 1991,

those that retained p ower were unable to supp ort their traditional working

class constituencies through so cial welfare and full employment p olicy Hu-

b er and Stephens forthcoming. In Scandinavia, Swedish and Norwegian left

parties are seen to b e \rapidly abandoning so cial demo cracy and embrac- 10

ing market lib eralism" Mo ene and Wallerstein 1993, 385. Through the

1980s and early 1990s, so cial demo crats in Norway and Sweden cut industry

subsidies, privatized state rms, deregulated nancial markets and generally

fo cused on price stabilityover full employment in economic p olicy Pontusson

1994, 35{38; Hub er and Stephens forthcoming; Mjset, Capp elen, Fagerb erg,

and Trany 1994, 67{70. Outside of Northern Europ e, left parties in France,

Australia, and New Zealand also turned to p olicies of deregulation, decen-

tralization, and privatization throughout the 1980s Ross and Jenson 1994;

Stilwell 1993; Massey 1995, ch. 3. This p olicy shift damaged the p olitical

exchange of public p olicy for wage mo deration. As a result we exp ect that

the negative e ect of lab or governments on wage growth will decrease in the

1980s and 1990s.

In sum, the erosion of union bargaining p ower, the rise of lo cal bar-

gaining, and the rightward shift of lab or government indicates a broad, but

uneven, \deinstitutionalization of lab or." In this new institutional context

lab or movements are less able to pursue to sectional interests in wage growth,

or more general interests in wage restraint.

In the institutional approach, the e ects of lab or supply and demand are

shap ed by the shift of institutional control away from lab or. The negative

e ect of unemploymentonwage growth is likely to b ecome more severe as

union aggressiveness in wage bargaining deteriorates. Lab or decline has also

reduced the of cost-of-living adjustments in wage contracts Western 1996a.

As a result, the negative in uence of in ation on wage growth should increase

in the 1980s and 1990s. Finally, unions are in a weaker p osition to assert

claims on the gains from technological improvements. Consequently, the

p ositive impact of pro ductivity growth on wages should also decrease in the 11

Table 2. Predicted e ects of market and institutional variables on real wage

growth, 18 OECD countries, 1966{1992.

Golden Age Change in Slow Growth

Regime E ect Regime

Unemployment

In ation

Pro ductivity Growth + 0

Bargaining Centralization + 0

Lab or Government + 0

Union Density + 0

recent p erio d.

The predicted e ects of the market and institutional variables are sum-

marized in Table 2. The rst column shows the e ects of all variables in the

golden age of lab or market p erformance. The market explanation suggests

that real wage growth dep ends negatively on unemployment and in ation,

but p ositively on pro ductivity growth. For the institutional explanation, cen-

tralized bargaining reduces wage growth, p erhaps in combination with lab or

government. Union density,however, raises, wage growth. Wage determina-

tion in the slow growth p erio d op erates di erently,however. We exp ect the

unemployment and in ation e ects to b ecome more negative. The p ositive

pro ductivity e ect should go to zero. All institutional e ects should also

movetowards zero, as the collective in uence of wage-earners declines.

A Model of Real Wage Growth and Labor Decline

Table 3 rep orts summary statistics for the indep endentvariables see Ap-

p endix 1 for data sources. A total of 483 country-years are used for analysis, 12

consisting of time series from 1966 to 1992 for all countries except Australia,

whose series ends in 1989. Standardized unemployment rates are used, where

available, to measure lab or demand. Unemployment is likely to b e endoge-

nous to wage growth, so unemployment e ects may b e in ated. Other re-

search on similar data sets suggests estimation with instrumental variables

has little e ect OECD 1997, 21, so we prefer the simple single-equation

mo del here. Bound, Jaeger, and Baker 1995 describ e the pitfalls of in-

strumental variables estimation. Like earlier research, price movements are

measured by the change in the in ation rate of the consumer price index

Layard et al. 1991; Volgy et al. 1996. Pro ductivity growth is given by

the p ercentage change in real gross domestic pro duct GDP p er employed

p erson.

Following Janoski, McGill and Tinsley 1997, all institutional measures

vary over time. Lab or government is measured by the prop ortion of cabinet

seats held by lab or, so cial demo cratic, so cialist, or communist parties. Union

density is measured by the total numb er of union memb ers including those

retired and unemployed as a p ercentage of all wage and salary earners plus

the unemployed. Time series are common for unionization and left govern-

ment measures, but unusual for indexes of bargaining centralization Crouch

1993, 14; cf. Hicks and Kenworthy 1997. We measure bargaining central-

ization with a four-p oint scale rep orted by Golden et al. 1997. High scores

indicate countries with national or sectoral bargaining that binds union af-

liates to no-strike agreements. Countries with decentralized rm-level or

industry bargaining score lowest. Centralization is scaled to vary b etween

0 and 1, so centralized settings like Norway and Sweden average close to 1,

while the United States and Canada are close to 0. It is sometimes argued 13

Table 3. Means of the indep endentvariables used in analysis of real wage growth

in 18 OECD countries, 1966{1992.

1 2 3 4 5 6

Australia 5.03 .15 1.85 .68 .40 51.25

Austria 2.44 -.03 3.16 .33 .72 64.07

Belgium 6.93 -.06 2.80 .51 .23 70.46

Canada 7.60 -.04 1.28 .11 .67 34.57

Denmark 5.62 -.16 1.43 .78 .43 79.52

Finland 4.40 -.09 3.07 .63 .48 78.24

France 6.13 .00 2.74 .33 .31 18.60

Germany 3.68 .02 2.62 .33 .40 40.37

Ireland 7.79 .04 3.11 .54 .15 59.49

Italy 10.24 -.07 3.84 .77 .18 53.89

Japan 1.96 -.18 4.33 .33 .00 31.08

Netherlands 5.87 -.03 2.09 .63 .18 37.79

New Zealand 3.00 -.10 .90 .62 .34 40.45

Norway 2.53 -.07 2.63 .91 .56 63.34

Sweden 2.45 -.10 1.66 .85 .73 86.62

Switzerland .51 .02 1.49 .33 .29 33.28

United Kingdom 6.68 -.04 2.01 .33 .35 50.07

United States 6.18 .05 .80 .07 .26 23.22

Note: Column headings are as follows: 1 unemployment; 2 in ation; 3 pro-

ductivity growth; 4 bargaining centralization; 5 lab or government; and 6

union density.

that decentralized but co ordinated industrial relations, like the Japanese, can

function similarly to centralized bargaining Soskice 1990. Weinvestigate

this idea with diagnostics that assess the sensitivity of results to data from

Japan, and other countries in the sample App endix 2.

The basic mo del for country i i =1;:::;18 at time t t = 1966;::: ;1992

is written:

w_ = b + b U + b y_ + b
_p + b L + b B + b D + e 1

it 0i 1 it 2 it 3 it 4 it 5 it 6 it it

where w_ is the annual p ercentage growth in real manufacturing wages, U 14

is the unemployment rate,y _ is real pro ductivity growth,
p _ is the annual

change in the in ation rate, L is lab or government, B is bargaining cen-

tralization, D is union density, and e is an error term. Subscripts on the

intercept indicate that cross-national di erences in average wage growth are

t with country-level dummyvariables. There is no residual auto correla-

tion with this mo del, but error variances di er across countries. We assume

that wage growth is conditionally normal, with di erentvariances for each

country. This mo del is estimated with maximum likeliho o d metho ds.

Theories of lab or movement decline suggest that institutional e ects may

change, but mo del 1 constrains all e ects to b e identical over time. Consis-

tent with the slow growth story, mo del 1 explains the decline in wage growth

with a general shift in the values of the indep endentvariables. Change in the

e ects of the variables can also b e treated as a parameter to b e estimated.

For this approach, the time series is divided into two regimes. At some p oint,

year k , the e ects of the indep endentvariables switch from the golden-age

wage regime to the slow-growth regime, describ ed in Table 2. For example,

supp ose we estimate that k = 1980, wewould exp ect that union density p osi-

tively e ects wage growth b efore 1980, but has little e ect after this year due

to the decline in union bargaining p ower. To study change p oints in wage

determination, we de ne a dummyvariable, D k , which scores zero for all

observations b efore year k , and one for observations from year k onwards.

The change-p oint mo del augments mo del 1 with the main e ect of D k 

and interaction e ects:

w_ = b + b U + b y_ + b
_p + b L + b C + b B +

it 0i 1 it 2 it 3 it 4 it 5 it 6 it

2

D k   b + b U + b y_ +

it 7 8 it 9 it

b
_p + b L + b C + b B +e

10 it 11 it 12 it 13 it it 15

Bayesian metho ds for mo del selection are used to decide b etween the constant-

e ects mo del of equation 1 and the change-p oint mo del 2 see West-

ern 1996b. Letting D k  shift over a range of p ossible break p oints, k =

1970;::: ;1990, we use the data to determine if and when a structural break

has o ccurred. This approach extends the historical time-series analysis of

Isaac and Grin 1989, by treating change p oints in statistical regimes as

quantities for estimation and inference.

The change-p oint mo del simpli es the historical record by assuming a

clear break in the wage growth pro cess whose timing is identical across coun-

tries. Although this assumption is certainly false, it o ers a useful simpli ca-

tion. With no clear break or heterogeneous timing of breaks across countries,

the analysis yields only weak evidence for a unique change-p oint. A more

realistic mo del would allowchange p oints to vary across countries. This sp ec-

i cation, however, would add over 100 new parameters, running the risk of

unidenti ed parameters and over- tting. The simpli ed approach of mo del

2 captures the main idea that lab or markets function di erently in the re-

cent p erio d of slow growth than in the golden age of the 1960s. If this is not

approximately true, evidence for the change-p oint will b e weak.

Results

The Bayes factor measures evidence for the change-p oint mo del see Ap-

p endix 3. Positive log Bayes factors show that the change-p oint mo del is

more probable than the constant-e ects mo del. Figure 1 rep orts a time se-

ries of the log Bayes factors for a range of alternative break p oints, k =

1970;::: ;1990. The year with the highest Bayes factor identi es the change-

p oint with highest p osterior probability. The data o er clear evidence for 16

a structural break in wage determination at 1975. This is shown by the

sharp p eak in the time series at this year. There are no other lo cal maxima,

and the change p oint mo del has much higher p osterior probability than the

constant-e ects mo del. This suggests that changes in wage determination

o ccur at roughly the same time for all countries|in 1975. The result is

unlikely to b e an artifact of volatility in 1975, as robust regression analy-

sis that downweights outliers yields substantively identical estimates. The

2

gure also rep orts a time series of R statistics from a naive OLS t for all

2

p ossible break p oints. The R statistics tell the same story as the Bayes

factors. Adding a dummyvariable for years 1975 and after, and interactions

with the dummyvariable, raises the p ercentage of explained variance by one-

quarter. Like other research that lo cates the end of lab or's golden age at the

mid-1970s Esping-Andersen 1990, 186; Crouch 1993, 291; Goldthorp e 1987,

the structural break identi ed in this analysis coincides with the recession

following the rst oil crisis.

Table 4 rep orts go o dness-of- t statistics for three mo dels: the constant-

e ects mo del, the change-p oint mo del, and a compromise mo del that includes

the p erio d dummyvariable, allowing a mean shift in wage growth after 1974.

2

All mo dels contain 17 country-level dummyvariables. Log-likeliho o d and R

statistics indicate that the mo dels with p erio d e ects t b est. The change-

p oint mo del ts signi cantly b etter than the p erio d e ect mo del, passing a

chi-square test at p<:01. The Bayes factor applies a more stringent test by

p enalizing highly parameterized mo dels Raftery 1995. The Bayesian crite-

rion also supp orts the change-p oint mo del. The mo del also ts well in the

qualitative sense of capturing substantively imp ortant patterns of variation.

Average wage growth in the OECD area was 3.25 p ercentage p oints slower 17 Log Bayes factor OLS R-Square OLS R-Square Log Bayes Factor 0102030 0.38 0.40 0.42 0.44 0.46 0.48

1970 1975 1980 1985 1990

Years

2

Figure 1. Log Bayes factors and OLS R statistics from a mo del of a structural

break in the determinants of real growth.

from 1975 to 1992 compared to 1966{1974 . The constant-e ects mo del as-

sumes the wage slowdown results only from changes in the indep endentvari-

ables. On this assumption, 60 of the slowdown in average wage growth is

explained. The change-p oint mo del, which allows the determinants of wage

growth to change over time, explains 80 of the wage slowdown in the p ost-

oil sho ck p erio d. The slow growth accountofwage stagnation based chie y

on rising in ation and unemployment and slow pro ductivity growth leaves

sizable unexplained variation.

Table 5 rep orts regression estimates from the change-p oint mo del, for

k = 1975. These estimates are conditional on the change-p oint, so con-

ventional unconditional standard errors and t statistics will b e optimistic.

Since the 1975 change-p ointisoverwhelmingly preferred by the data this

bias is extremely small. Of all the change-p oint mo dels, k = 1975 has 18

Table 4. Go o dness of t statistics for three mo dels of wage growth.

Mo del

1 2 3

Mo del Description M M + P M + P + I

Numb er of co ecients 24 25 31

2

R .37 .45 .47

Log Likeliho o d -1060 -1029 -1010

Log Bayes Factor - 28.9 40.0

a

Explained Decline  60 70 80

Note: M =main e ects, P = p erio d e ect, and I =p erio d interaction e ects. The

log Bayes factor compares mo dels 2 and 3 to the baseline mo del 1.

a

Percentage of the decline in the average rate of wage growth b etween 1966{1974

and 1975{1992 explained by the mo del.

99 p osterior probability; averaging over p ossible change-p oints for uncon-

ditional inference thus yields essentially identical results. The rst column

of the table describ es wage growth b etween 1966 and 1974. These results

provide mo dest evidence for the in uence of market forces. Consistent with

theory,wage growth is negatively asso ciated with in ation and unemploy-

ment. However, the co ecients are small with large variance. In the late

1960s and early 1970s unemployment rates were generally low, showing little

variation. These trends combined with the small unemployment co ecient

suggests lab or demand exerted little in uence on real wage growth b efore the

oil crisis. While unemploymentwas low, the end of the golden age featured

several large in ationary sho cks in the late 1960s. The in ation e ect indi-

cates that the real value of wages were shielded from steep price rises at this

time. Pro ductivity growth provides the most p owerful economic e ect of the 19

golden-age wage regime. The co ecient is large and statistically signi cant.

This estimate usefully distinguishes trends in wage growth in Europ e and

North America. Pro ductivity grew at ab out 4 annually in Italy, Germany,

and France b efore 1975, contributing more than p oint to real wage growth.

U.S. pro ductivity growth averaged four- fths of a p ercentage p oint b efore

the oil sho ck, generating less than one-quarter of a p ointinwage growth.

While market e ects are generally weak, lab or movements strongly in u-

enced wages b efore 1975. Contrary to theory,wages grew faster in countries

with centralized bargaining and lab or governments. The p ositive e ect of

lab or government is large and signi cantatconventional levels. These esti-

mates weaken the claim that centralized representation of lab or movements

cultivated a general interest in wage restraint. One p ossible interpretation

places these results in historical p ersp ective. In the late 1960s and early

1970s, lab or governments supp orted their working class electorates by main-

taining wage standards. During this time, lab or parties backed central union

wage p olicies for low-payworkers. So cial demo crats also intervened less

in industrial relations compared to conservatives who more commonly pre-

empted collective bargaining with wage freezes Flanagan et al. 1983 provide

evidence for the Netherlands and Denmark. In addition, so cial demo crats

allowed large public sector wage increases which spilled over into the economy

as a whole. In short, solidaristic p olicies to raise wages among low-paywork-

ers and general supp ort for wage standards may partly explain the p ositive

wage e ects of lab or government. Following other research [Alvarez et al.

1991; Hicks 1994], we also exp erimented with interactions b etween bargaining

level and lab or government, but these were not signi cant. Supp orting the

power resources theory of union wage e ects, wages also grew faster in coun- 20

Table 5. Regression results in a mo del of real wage growth in 18 OECD countries,

1966{1992. Absolute t-statistics in parentheses.

Main Perio d Post-1975

E ects Interactions Net E ects

1 2 1 + 2

Intercept -.23 1.10 .87

.19 1.73 .91

Unemployment -.08 -.17 -.26

.80 1.88 5.40

In ation -.01 -.25 -.25

.12 3.15 4.72

Pro ductivity Growth .28 -.25 .03

4.24 2.69 .42

Bargaining Centralization .68 -1.64 -.96

1.17 2.38 1.90

Lab or Government .80 -1.06 -.26

1.94 2.13 .86

Union Density .40 -.26 .13

1.99 2.14 .95

Note: The second column rep orts interactions with the p erio d dummyvariable

indicating observations from 1975 to 1992. The p erio d e ect is the intercept term

in column one and the co ecient for the p erio d dummyvariable in column two.

Union density co ecients have b een multiplied by 10. Co ecients for country-

level dummies have b een suppressed. In column 3, variances for the sum of main

e ects, b , and interactions, b equals V b + V b + 2cov b ;b .

M I M I M I 21

tries with extensive unionization. A 50 p oint di erence in union density|the

average di erence b etween Sweden and the United States|is asso ciated with

a 2 p ercentage p oint di erence in wage growth in the golden-age wage regime.

The wage growth pro cess changed markedly after 1975. Interactions de-

scrib e the change in e ects after the oil sho ck column 2, Table 5. The

sum of the interaction and main e ects provide the net e ects for the slow-

growth p erio d column 3, Table 5. Supp orting the idea that wages are more

vulnerable to unemployment when union bargaining p ower is weak, the net

negative impact of unemployment tripled in the slow growth p erio d. The

increasingly negative e ect comp ounds the rise in joblessness, suggesting the

extremely large in uence of unemployment on recent Europ ean wage trends.

In ationary sho cks also checked wage growth after 1975. This result un-

derlines the diminishing in uence of cost-of-living adjustments in union con-

tracts. Despite these estimates, market mechanisms did not unambiguously

expand their in uence in the slow growth p erio d. The e ect of pro ductivity

growth moves towards zero after 1975. Although pro ductivity growth drove

pay rises b efore the oil crisis, pro ductivity trends are weakly related to wage

movements afterwards. Before 1975, years of p ositive pro ductivity growth

coincided with p ositivewage growth more than 90 of the time. After 1975,

p ositivewage growth accompanied pro ductivity increases just 60 of the

time. This suggests that workers and their representatives less successfully

asserted claims on pro ductivity improvements over the last two decades.

Results for the institutional e ects also show signi cant di erences with

the pre-oil sho ck p erio d. Unexp ectedly, the bargaining centralization e ect

turns signi cantly negative in the slow growth p erio d. Despite the growth of

lo cal bargaining, this estimate provides evidence of centrally-organized wage 22

restraint. This estimate may result from the increasing imp ortance of cen-

tralized bargaining during recessions or in contexts of intensi ed economic

volatility.Thus case studies show that centralized bargaining pro duced wage

restraint in Austria in the 1980s and in many of the small Europ ean coun-

tries in the late 1970s Katzenstein 1985; Scharpf 1991; Flanagan et al. 1983.

Whatever the interpretation, there app ears signi cant evidence for institu-

tional continuity in collective bargaining and little supp ort for a universal

deregulation of OECD lab or markets.

The left government e ect for the 1980s changes similarly to the bargain-

ing level e ect. The data show that wage growth was relatively slow under

lab or and so cial demo cratic government. Again, this result contradicts his-

torical evidence that lab or movementswere increasingly unable to pursue

general interests in wage restraint in the 1980s and 1990s. Still, similar to

other research O'Connell 1994, the net e ect for the p ost-oil sho ck p erio d

is small, and not statistically di erent from zero.

Finally, the p ositive e ect of union density that we nd b efore 1975 is

close to zero. Although unions in several cases maintained their organi-

zational strength, this suggests their bargaining p ower was severely weak-

ened. Lo cal unions were less able to pursue their sectional interests in higher

wages, and the relationship b etween union density and wage growth at-

tened. In countries where unions lost memb ership, as in the United States or

the United Kingdom, the impact of deunionization seems esp ecially severe.

In these cases, the capacity of unions to protect living standards su ered a

double blow: declining organization reduced the reach of union wages, while

diminished bargaining p ower reduced wages increases obtained by unionized

workers. These results are consistent with U.S. studies show the decline of 23

union memb ership and the increasing incidence of concession bargaining in

the 1980s Farb er 1989; Mitchell 1993.

Avariety of other mo dels were also studied for this analysis. In addition to

the results rep orted we also examined the e ects of trade, economic growth,

interactions b etween economic and institutional variables, nonlinearities in

centralization e ects, and a range of alternative institutional variables. The

results from the simple mo del rep orted here are the most robust and among

the strongest, but similar mo dels yield similar conclusions. A systematic

survey of alternative mo dels rep orts intervals of co ecients, obtained when

estimating all p ossible subsets of the indep endentvariables. Sensitivityof

the results to information from individual countries was also assessed with

a jackknife analysis. Both typ es of sensitivity analyses are rep orted in Ap-

p endix 2, demonstrating the robustness of the rep orted ndings to outliers

and mo del assumptions.

Discussion

This analysis provides novel evidence of a structural break in the pro cess of

wage determination some time in the mid-1970s in the advanced capitalist

lab or markets. Before the break, wages were insulated from the e ects of

unemployment and in ation, while pro ductivity growth assisted a continuous

rise in earnings. Union organization, bargaining centralization, and lab or

government are all asso ciated with rising wages in the golden age. The oil

crisis initiated a novel typ e of recession that set all the advanced capitalist

lab or markets on a new path of development. Not only did the values of

key variables change in a way that hurt wages in the mid-1970s; the causal

pro cess of wage determination also shifted. In the the slow growth era, wage- 24

earners were vulnerable to rising unemployment and in ation and less able

to share in gains of technological progress. The institutional determinants of

wage growth also changed. The p ower of union density to raise wages was

substantially curtailed. The p ower of state actors to a ect wages was also

weakened. In contrast to claims of the dissolution of centralized bargaining,

however, there is strong evidence of centralized wage restraint in the 1980s

and 1990s.

Metho dological, substantive, and theoretical conclusions can b e drawn

from this study. First, this pap er contributes to the metho dology of institu-

tional analysis. Recently, comparative researchers have b een urged to view

institutions dynamically,aschanging over time Janoski et al. 1997. In some

cases, however, indexes of formal institutional features neglect changes in in-

formal features, or changes in context which shap e institutional e ects. Such

developments suggest institutional e ects maychange over time, even when

institutions are measured longitudinally Isaac and Grin 1989. We treat

this as a problem of parameter estimation, in which the sample data help

decide the most likely change-p oint in institutional e ects. In this approach,

dynamic pro cesses generate not just a change in the value of institutional

variables, but also a change in institutional e ects.

From a substantive viewp oint, the general pattern of results indicates a

signi cant transfer of risk in capitalist economies from employers and the

state to wage earners. The emergence of unemployment and in ation e ects

suggests that living standards are increasingly sensitive to market uctua-

tions. The declining e ect of union density indicates that collective action

in the lab or market has b ecome less e ective for maintaining wages. The

p olitical sources of wage growth were also ero ded. Only the increase in the 25

bargaining centralization e ect provides evidence of resilient collective control

over lab or market outcomes. These ndings parallel other research. Studies

of the U.S. lab or market nd growing instability of employment and earnings

Bernhardt, Morris, Handco ck and Scott 1997; Gottschalk and Mott 1994.

Moreover, income supp ort in the United States and Europ e increasingly ties

so cial b ene ts to tougher conditions for job searching and training McFate

1995. Changes in the pro cess of wage determination thus contributes to a

more general trend to the receding role of so cial protection and the growing

role of markets in the allo cation of living standards in the advanced capitalist

countries.

Most generally, the analysis also suggests that lab or markets are deeply

p olitical forums for economic allo cation. By this we mean that lab or market

outcomes are shap ed by the surrounding balance of p ower b etween owners

and wage earners. This idea is illustrated most clearly with evidence for in-

stitutional e ects. More fundamentally p erhaps, the p ower relations b etween

owners and workers also app ear to a ect the relationships among economic

variables. Thus the link b etween pro ductivity growth and wages is less a

necessity of comp etitive markets and more a contingent fact of the capacity

of wage earners to assert claims on the dividends of technical progress. Sim-

ilarly, the negative e ects of unemployment and in ation on wage growth

are also historically variable, shifting to the disadvantage of wage-earners,

when the institutional p osition of organized lab or movements is weakened.

From this p ersp ective, the OECD wage slowdown seems proximately and

partly caused by the economics of the slow-growth era, but fundamentally

dep endent on the declining p ower of organized lab or movements. 26

Appendix 1. Data Sources

Real Wage Growth Annual p ercentage changes in real hourly earnings in

manufacturing are mostly taken from OECD 1996, but national sources

have b een used in some cases, owing to missing data. Australian data have

b een supplemented by the weekly award average weekly earnings series for

male wage and salary earners Australian Bureau of Statistics 1988, 294;

1989, 190; 1992, 203. Austrian data are from the monthly earnings in min-

ing and manufacturing series of OECD 1993. Danish and Swedish wage

data have b een compiled from tables of wages in mining and manufactur-

ing published by the Nordic Council of Ministers and the Nordic Statistical

Secretariat 1974, 185; 1983, 239; 1991, 253; 1994, p. 238. Swedish data

were adjusted to account for the inclusion of overtime and holidaypayin

the earliest table. Finally, the Dutch series was completed with data for

the average hourly earnings of male manufacturing sector workers in tables

published by the Netherlands Central Bureau of Statistics 1970, 371; 1971,

313; 1975, 305; 1978, 324; 1980, 344. Where national sources were used to

complete OECD series, data were smo othed to eliminate discontinuities and

overlapping national and OECD series were compared to ensure comparabil-

ity.

Unemployment Standardized unemployment rates have b een used where

available. Unstandardized gures based on the numb er of unemployed as a

p ercentage of the civilian lab or force have b een used for Austria, Denmark,

Ireland, and Switzerland. All data come from OECD 1996, except for New

Zealand and Switzerland which also use data from Layard et al. 1991, 526{

29.

Productivity Growth Annual p ercentage changes in real GDP p er p erson em-

ployed are taken from OECD 1996.

In ation Annual p ercentage changes in the consumer price index are taken

from OECD 1996.

Labor Government To obtain the p ercentage of cabinet seats held by lab or

parties, cabinet representation was co ded for every quarter. Where there was 27

achange in cabinet representation, the longest-serving cabinet in the quarter

was co ded. Annual averages were then take from the quarterly series. Infor-

mation ab out the party comp osition of cabinets is taken from Woldendorp,

Keman, and Budge 1993. These data were up dated with tables rep orted

in Ko ole and Mair 1994.

Bargaining Centralization A four-p oint scale describing the highest level at

whichwages are determined: 1 plant-level wage-setting, 2 industry-level

wage setting, 3 sectoral wage-setting without sanctions, and 4 sectoral

wage-setting with sanctions i.e., wage bargains include no-strike clauses.

Golden et al. 1997 rep ort time-series data for all countries except Ireland

and New Zealand. We supplied co des or these two countries, using Hince

1986, Hince and Vranken 1991, and Gunnigle, McMahon, and Fitzgerald

1995.

Union Density The p ercentage of workers who are union memb ers. The

density series combines from the gross density series of Visser 1992b, 1996.

Some missing data were interp olated. Discontinuities in the series owing to

data discrepancies, were smo othed.

Appendix 2. Sensitivity Analysis

Sensitivity of the results to the data and the mo del assumptions are studied

in Table A.1. Sensitivity to the data is studied with a typ e of jackknife

that estimates 18 sets of regression co ecients calculated from reduced data

sets with a single country omitted. The main results are robust to this

metho d and nearly all signi cant co ecients retain their signs in the jackknife

analysis.

Because the mo del is not known with certainty, results are sensitiveto

the choice of indep endentvariables. We assessed this sensitivity with an

\extreme b ounds analysis" Leamer 1983. This involved re-estimating the

mo del using all p ossible subsets of the economic variables. The highest and

lowest co ecient estimates form intervals describing mo del uncertaintyTa-

ble A.1. The results are robust to changes in the mo dels, and sp eci cation

uncertainty do es not substantively change the inferences rep orted ab ove. 28

Table A.1. Cross-validation and extreme b ounds sensitivity analyses of real wage

growth analysis, 18 OECD countries, 1966{1992.

Jackknife Extreme Bounds

Intercept [-1.00, .84] [-.36, 3.24]

Unemployment [-.16, -.05] [-.22, -.03]

In ation [-.06, .02] [-.05, .04]

Pro ductivity Growth [.18, .32] [ .26, .37]

Bargaining Centralization [.14, 1.23] [ .26, .98]

Lab or Government [.34, 1.06] [ .26, 1.01]

Union Density [.26, .61] [.18, .51]

D k  [.78, 1.33] [-3.38, 1.10]

D k Unemployment [-.22, -.13] [-.17, -.02]

D k In ation [-.27, -.21] [-.26, -.15]

D k Pro ductivity growth [-.28, -.15] [-.36, -.18]

D k Bargaining Centralization [-2.25, -1.09] [-2.34, -1.24]

D k Lab or Government [-1.38, -.51] [-1.44, -.42]

D k Union Density [-.39, -.16] [-.46, -.14]

Note: Union density main e ect and interaction e ect have b een multiplied by

ten. 29

Appendix 3. Bayesian Analysis of the Structural Break

From the Bayesian p ersp ective, the change-p ointinwage determination is

identi ed by tting a range of mo dels with break p oints, k = 1970;::: ;1990,

and calculating their p osterior probabilities. The Bayes factors, B , express

k 0

the p osterior probability of the change-p oint mo dels, M , as a ratio of the

k

p osterior probability of the constant-e ects mo del with no break p oint, M .

0

With di use prior information the log Bayes factor, B , can b e approxi-

ij

mated using quantities from maximum likeliho o d estimation,

log B = I I

ij i j

where

^

I =2 p =2 + log jV  j=2+`

i i i

^

where p is the numb er of co ecients in M , V   is the covariance matrix

i i i

of the maximum likeliho o d estimates of the co ecients, , and ` is the

i i

maximized log likeliho o d. Log Bayes factors in Figure 1 were based on this

approximation.

References

Alvarez, R. Michael, Geo rey Garrett, and Peter Lange. 1991. \Government

Partisanship, Lab or Organization, and Macro economic Performance."

American Political ScienceReview 85:539{556.

Australian Bureau of Statistics. 1988, 1989, 1992. Ocial Year Book of the

Commonwealth of Australia. Canb erra, Australia: Australian Bureau

of Statistics.

Baglioni, Guido and Colin Crouch eds. 1990. European Industrial Rela-

tions: The Chal lenge of Flexibility. Newbury Park: Sage.

Bernhardt, Annette, Martina Morris, Mark Handco ck, and Marc Scott. 1997.

\Job Instability and Wage Inequality: Preliminary Results from Two

NLS Cohorts." Pap er presented at the Conference on lab or Market

Inequality. March, 1997. Madison, Wisconsin. 30

Blanch ower, David G. and Richard B. Freeman. 1992. \Unionism in the

United States and the Other Advanced OECD Countries." Industrial

Relations 31:56{79.

Bound, John, David A. Jaeger, and Regina M. Baker. 1995. \Problems

With Instrumental Variables Estimation When the Correlation Be-

tween the Instruments and the Endogenous Explanatory Variables is

Weak." Journal of the American Statistical Association 90:443{450.

Calmfors, Lars and Drill, John. 1988. \Centralisation of Wage Bargaining

and Macro economic Performance." Economic Policy 6:13{62.

Chan-Lee, James, David T. Co e and Menahem Prywes. 1987. \Micro e-

conomic Changes in Macro economic Wage Disin ation in the 1980s."

OECD Economic Studies 8:121{158.

Cohn, Samuel. 1993. When Strikes Make Sense|and Why: Lessons from

ThirdRepublic French Coal Miners. New York: Plenum.

Crouch, Colin. 1985. \Conditions for Trade Union Wage Restraint." Pp. 105{

139 in The Politics of In ation and Economic Stagnation: Theoretical

Approaches and International Case Studies, edited by Leon Lindb erg

and Charles Maier. Washington, DC: Bro okings.

Crouch, Colin. 1993. Industrial Relations and European State Traditions.

New York: Clarendon.

Epstein, Gerald and Juliet B. Schor. 1992. \Structural Determinants and

Economic E ects of Capital Controls in OECD Countries." Pp. 136{

161 in Financial Openness and National Autonomy, edited byTariq

Banuri and Juliet B. Schor. Oxford: Clarendon.

Esping-Andersen, Gsta. 1990. Three Worlds of Welfare Capitalism. Prince-

ton, NJ: Princeton University Press.

Farb er, Henry S. 1990. \The Decline of Unionization in the United States:

What Can b e Learned From Recent Exp erience?" Journal of Labor

Economics 8:S75{S105. 31

Flanagan, Rob ert J., David W. Soskice, and Lloyd Ulman. 1983. Unionism,

Economic Stabilization, and Incomes Policies: European Experience.

Washington: Bro okings.

Freeman, Richard B. 1995. \The Limits of Wage Flexibility to Curing Un-

employment." OxfordReview of Economic Policy 11:63{72.

Friedman, Milton. 1968. \The Role of Monetary Policy." American Eco-

nomic Review 58:1{17.

Gershuny, Jonathan. 1983. Social Innovation and the Division of Labor.

Oxford: Oxford University Press.

Glyn, Andrew. 1995. \So cial Demo cracy and Full Employment." New Left

Review 211:33{55.

Golden, Miriam, Peter Lange, and Michael Wallerstein. 1993. \Trends

in Collective Bargaining and Industrial Relations in Non-Corp oratist

Countries: A Preliminary Rep ort." Pap er presented at the Annual

Meetings of the American Political Science Asso ciation. Washington,

DC.

Golden, Miriam, Peter Lange, and Michael Wallerstein. 1997. \Master Co de-

b o ok for NSF Data on 16 Countries." Typ escript co deb o ok. Los An-

geles: UCLA.

Goldthorp e, John H. 1984. \The End of Convergence: Corp oratist and Dual-

ist Tendencies in Mo dern Western So cieties." Pp. 315{343 in Order and

Con ict in Contemporary Capitalism: Studies in the Political Economy

of Western European Nations, edited by John H. Goldthorp e. Oxford:

Clarendon Press.

Goldthorp e, John H. 1987. \Problems of Political Economy After the Post-

war Perio d." Pp. 363{408 in Changing Boundaries of the Political:

Essays on the Evolving Balancebetween the State and Society, Public

and Private in Europe, edited by Charles S. Maier. Cambridge: Cam-

bridge University Press. 32

Gottschalk, Peter, and Rob ert Mott. 1994. \The Growth of Earnings

Instability in the U.S. Lab or Market." Brooking Papers on Economic

Activity 2:217{272.

Gunnigle, Patrick, Gerard McMahon and Gerard Fitzgerald. 1995. Industrial

Relations in Ireland: Theory and Practice. Dublin:Gill and McMillan.

Harrison, Bennett and Barry Bluestone. 1988. The Great U-Turn: Corporate

Restructuring and the Polarizing of America. New York: Basic Bo oks.

Headey, Bruce. 1970. \Trade Unions and National Wages Policy." Journal

of Politics 32:407{39.

Hibbs, Douglas A. 1987. The Political Economy of Industrial Democracies.

Cambridge, MA: Harvard University Press.

Hicks, Alexander M. 1994. \The So cial Demo cratic Corp oratist Mo del of

Economic Performance in the Short- and Medium-run Persp ective."

Pp. 189{217 in The Comparative Political Economy of the Welfare

State, edited by Thomas Janoski and Alexander Hicks. New York:

Cambridge University Press.

Hicks, Alexander and Lane Kenworthy. 1997. \Co op eration and Political

Economic Performance in Auent Demo cratic Capitalism." Unpub-

lished manuscript. Emory University.

Hicks, John. 1963. The Theory of Wages. London: Macmillan.

Hince, Kevin 1986. \Wage xing in a p erio d of change: The New Zealand

Case." International Labour Review 125:463{472.

Hince, Kevin and Martin Vranken. 1991. \ A Controversial Reform of New

Zealand lab our Law: The Employment Contracts Act of 1991." Inter-

national Labour Review 130.

Hub er, Evelyne and John D. Stephens. Forthcoming. \Internationalization

and the So cial Demo cratic Mo del: Crisis and Future Prosp ects." Com-

parative Political Studies. 33

Isaac, Larry W. and Larry J. Grin. 1989. \Ahistoricism in Time-Series

Analysis of Historical Pro cess: Critique, Redirection, and Illustration

from U.S. Lab or History." American Sociological Review 54:873{890.

Janoski, Thomas, Christa McGill, and Vanessa Tinsley. 1997. \Making Insti-

tutions Dynamic in Cross-National Research: Time-Space Distancing

in Explaining Unemployment." Comparative Social Research 16:227{

68.

Katz, Harry C. 1993. \The Decentralization of Collective Bargaining: A

Literature Review and Comaprative Analysis." Industrial and Labor

Relations Review 47:3{22.

Katz, Lawrence F., Gary W. Loveman, David G. Blanch ower. 1995. \A

Comparison of Changes in the Structure of Wages in Four OECD Coun-

tries." Pp. 25{66 in Di erences and Changes in Wage Structures,

edited by Richard B. Freeman and Lawrence F. Katz. Chicago: Uni-

versity of Chicago Press.

Katzenstein, Peter J. 1985. Smal l States in World Markets. Ithaca, NY:

Cornell University Press.

Keynes, John Maynard. [1935] 1964. The General Theory of Employment,

Interest, and Money. San Diego: Harcourt Brace, Jovanovich.

Ko ole, Ruud and Peter Mair. 1994. Political Data Yearbook, 1994: European

Journal of Political Research 263/4.

Korpi, Walter. 1983. The Democratic Class Struggle. Boston: Routledge

and Kegan Paul.

Korpi, Walter and Michael Shalev. 1979. \Strikes, Industrial Relations and

Class Con icts in Capitalist So cieties." British Journal of Sociology

30:164{87.

Lange, Peter, Michael Wallerstein, and Miriam Golden. 1995. \The End of

Corp oratism? Wage Setting in the Germanic and Nordic Countries."

Pp. 76{100 in The Workers of Nations: Industrial Relations in a Global 34

Economy, edited by Sanford Jacoby. New York: Oxford University

Press.

Layard, Richard, Stephen Nickell, and Richard Jackman. 1991. Unemploy-

ment: Macroeconomic Performance and the Labour Market. Oxford:

Oxford University Press.

Leamer, Edward E. 1983. \Let's Take the Con out of Econometrics." Amer-

ican Economic Review 23:31{43.

Marsh, David. 1992. The New Politics of British Trade Unionism: Union

Power and the Thatcher Legacy. Ithaca, NY: ILR Press.

Massey,Patrick. 1995. New Zealand: Market Liberalization in a Developed

Economy. New York: St. Martin's.

McFate, Katherine. 1995. \Tramp olines, Safgety Nets, or Free Fall? lab or

Market Policies and So cial Assistance in the 1980s." Pp. 631{664 in

Poverty, Inequality and the FutureofSocial Policy: Western States in

the New World Order, edited by Katherine McFate, Roger Lawson and

William Julius Wilson. New York: Russell Sage Foundation.



Mjset, Lars, Adne Capp elen, Jan Fagerb erg, and Bent Sofus Trany. 1994.

\Norway: Changing the Mo del." Pp. 55-76 in Mapping the West Eu-

ropean Left, edited byPerry Anderson and Patrick Camiller. London:

Verso.

Mitchell, Daniel J.B. 1993. \Keynesian, Old Keynesian, and New Keynesian

Wage Nominalism." Industrial Relations 32:1{29.

Mo ene, Karl Ove and Michael Wallerstein. 1993. \The Decline of So cial

Demo cracy." Pp. 385{403 The Economic Development of Denmark

and Norway since 1870, edited by Karl Gunnar Persson. Bro ok eld

VT: Elgar.

Netherlands Central Bureau of Statistics. 1970, 1971, 1975, 1978, 1980. Sta-

tistical Yearbook of the Netherlands. The Hague: Netherlands Central

Bureau of Statistics. 35

Nordic Council of Ministers and the Nordic Statistical Secretariat. 1974,

1983, 1991, 1994. Yearbook of Nordic Statistics. Sto ckholm: Nordic

Council of Ministers.

O'Connell, Phillip J. 1994. \National Variation in the Fortunes of Lab or: A

Po oled and Cross-Sectional Analysis of the Impact of Economic Crisis

in the Advanced Capitalist Nations." Pp. 218{244 in The Comparative

Political Economy of the Welfare State, edited by Thomas Janoski and

Alexander M. Hicks. New York, NY: Cambridge University Press.

OECD. 1993. Main Economic Indicators, Historical Statistics: Prices, Labour

and Wages.Paris: OECD.

OECD. 1994. The OECD Jobs Study: Part II { The Adjustment Potential

of the Labor Market. Paris: OECD.

OECD. 1996. Historical Statistics 1960{1994 on Diskette.Paris: OECD.

OECD. 1997. Employment Outlook. Paris: OECD.

Olson, Mancur. 1982. The Rise and Decline of Nations: Economic Growth,

Stag ation, and Social Rigidities. New Haven, CT: Yale University

Press.

Piven, Frances Fox. 1991. \The Decline of Lab or Parties: An Overview."

Pp. 1{19 in Labor Parties in Postindustrial Societies, edited byFrances

Fox Piven. New York: Oxford University Press.

Pontusson, Jonas. 1994. \Sweden: After the Golden Age." Pp. 23{54 in

Mapping the West European Left, edited byPerry Anderson and Patrick

Camiller. London: Verso.

Pontusson. 1995. \Explaining the Decline of Europ ean So cial Demo cracy:

The Role of Structural Economic Change." World Politics 47:495-533.

Raftery, Adrian. 1995. \Bayesian Mo del Selection in So cial Research." So-

ciological Methodology 25:111{164. 36

Ross, George and Jane Jenson. 1994. \France: Triumph and Tragedy." Pp.

158{88 in Mapping the West European Left, edited byPerry Anderson

and Patrick Camiller. London: Verso.

Rubin, Beth A. 1986. \Class Struggle American Style: Unions, Strikes, and

Wages." American Sociological Review 51:618{31.

Scharpf, Fritz. 1991. Crisis and Choice in European Social Democracy .

Ithaca, NY: Cornell University Press.

Shalev, Michael. 1992. \The Resurgence of Lab our Quiescence." Pp. 102{

132 in The FutureofLabour Movements , edited by M. Regini. London,

England: Sage.

Soskice, David. 1990. \Wage Determination: The Changing Role of In-

stitutions in Advanced Industrialized Countries." OxfordReview of

Economic Policy 6:1{23.

Stilwell, Frank. 1993. \Wages Policy and the Accord." Pp. 65{84 in

The Australian Economy Under Labor, edited by Greg Mahoney. St.

Leonard, Australia: Allen and Unwin.

Streeck, Wolfgang. 1993. \The Rise and Decline of Neo corp oratism." Pp. 80{

101 in Labor and an Integrated Europe, edited by Lloyd Ulman, Barry

Eichengreen, and William Dickens. Washington DC: Bro okings.

Thelen, Kathleen. 1993. \West Europ ean Lab or in Transition: Sweden and

Germany Compared." World Politics 46:23{49.

Visser, Jelle. 1992a. \The Strength of Union Movements in Advanced Cap-

italist Demo cracies: So cial and Organizational Variation." Pp. 17{52

in The FutureofLabor Movements, edited by Marino Regini. London:

Sage.

Visser, Jelle. 1992b. \Trade Union Memb ership Database." Unpublished

data le. Amsterdam: Department of So ciology, University of Amster-

dam. 37

Visser, Jelle. 1996. \Unionisation Trends Revisited." Center for Research

of Europ ean So cieties and Industrial Relations research pap er 1996/2.

Amsterdam, the Netherlands.

Volgy, Thomas J., John E. Schwarz, and Lawrence E. Inwalle. 1996. \In

Search of Economic Well-Being: Worker Power and the E ects of Pro-

ductivity, In ation, and Unemployment and Global Trade on Wages in

Ten Wealthy Countries." American Journal of Political Science 1233{

1252.

Weiler, Paul C. 1990. Governing the Workplace: the FutureofLabor and

Employment Law. Cambridge: Harvard University Press.

Western, Bruce. 1996a. \RecentWage Trends in 14 OECD Countries."

Pap er presented at the meetings of Research Committee 28 of the In-

ternational So ciological Asso ciation. Ann Arb or, MI.

Western, Bruce. 1996b. \Vague Theory and Mo del Uncertainty in Macroso-

ciology." Sociological Methodology 26:165{92.

Western, Bruce. 1997. Between Class and Market: Postwar Unionization in

the Capitalist Democracies. Princeton, NJ: Princeton University Press.

Woldendorp, Jaap, Hans Keman and Ian Budge. 1993. \Parties and Govern-

ments in Industrialized Parliamentary Demo cracies." European Journal

of Political Research 24:1{120.

Wo o d, Adrian. 1994. North-South Trade, Employment, and Inequality:

Changing Fortunes in a Skil l-Driven World. Oxford: Clarendon. 38